The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLVIII-4/W6-2022
https://doi.org/10.5194/isprs-archives-XLVIII-4-W6-2022-457-2023
https://doi.org/10.5194/isprs-archives-XLVIII-4-W6-2022-457-2023
14 Feb 2023
 | 14 Feb 2023

ASSESSING AND MODELLING URBAN HEAT ISLAND IN BAGUIO CITY USING LANDSAT IMAGERY AND MACHINE LEARNING

D. C. D. M. Vergara, A. C. Blanco, J. J. S. Marciano Jr., S. F. Meneses III, N. J. B. Borlongan, and A. J. C. Sabuito

Keywords: Urban Heat Island, Regression Modelling, UTFVI, LST, NDVI, NDBI

Abstract. The Urban Heat Island (UHI) is a phenomenon where an urban area experiences higher temperatures than its surroundings. A commonly observed phenomenon worldwide and is one of the serious environmental problems related to urbanization. This paper assessed the past and current state of UHI in Baguio City, the Summer Capital of the Philippines. Land Surface Temperature (LST) layers were generated from Landsat images (March 25, 2019, and March 09, 2022) using the Project GUHeat Toolbox and then used to calculate the Urban Thermal Field Variance Index (UTFVI). The study found out that the UHI has intensified in the past three years. In contrast LST in March 2022 was generally lower than that in March 2019, most likely due to differences in weather conditions. This implies that while it is important to examine the spatiotemporal variations of LST, it is critical that UHI indices are also examined. Random Forest regression was used to examine the UHI indices such as Normalized Difference Built-up Index (NDBI) and Normalized Difference Vegetation Index (NDVI). The explanatory variables used in modelling are (1) NDBI (2) NDVI (3) combination of NDBI and NDVI. The performance of the models is evaluated with Mean Squared Error (MSE) and R-squared (R2). Using NDBI or NDVI alone yielded a less satisfactory model. The combination of NDBI and NDVI resulted in a good prediction of UHI with R2=0.89 and MSE=0.006.